On a learning rate factor and extent of ordering in basic self-organizing maps (Nonlinear Analysis and Convex Analysis)

Abstract

We deal with an essential mathematical model for self-organizing maps referred to as Kohonen type algorithm. Self-organizing map algorithm has been widely used as a useful tool in many practical problems on extensive fields. By repeating learning, some model functions in self-organizing maps have some properties of mathematical interest such as regularity between the nodes and their values. We mainly describe a learning late factor and extent of ordering in essential self-organizing maps with one-dimensionally indexed array. We present some numerical examples and make a comparison of the degree of converging for some learning processes in a basic self-organizing map.